When you assign a weight to a child, keep in mind that the weight you use reflects the importance of a child to its parent topic. The matched documents are ranked by importance to the search; thus, your selection results are directly affected by the weights you assign. If you change a weight, your selection results change as well.
Example:
The topic boeing-people includes three weighted children, binder, shrontz, and woodard, as shown below.
These subtopics are assigned various weights, as follows: the child binder is assigned a weight of 0.80, since this child is considered to be the most important of the three. The subtopic hitsman is assigned a “median” weight of 0.50, since this child is reasonably important with respect to the other two children. The subtopic johnson is assigned a low weight of 0.30, since this child is considered to be the least important with respect to the other children.
When the topic boeing-people is used for a search, the Verity search engine assumes that if the phrase “Paul Binder” is located within a document, there is a high probability that the document is relevant to a search that uses the topic boeing-people. Documents that contain the phrase “Frank Shrontz” are reasonably relevant to this search; documents that contain the phrase “Ron Woodard” are the least relevant.
Because the topic boeing-people has been assigned the ACCRUE operator, the documents displayed at the top of the results list are those that contain the greatest number of children; therefore all documents with references to all three people are given the most importance. Documents that contain only one name will be selected in an order that reflects the weights of each child. Thus, because the binder child has the highest weight, documents that include only one individual are ranked by those that refer to Paul Binder first, followed by Frank Shrontz, and finally Ron Woodard.
